Am. J. Hum. Genet.

73:627631, 2003

Report

ACTN3 Genotype Is Associated with Human Elite Athletic Performance

Nan Yang,1 Daniel G. MacArthur,1,2 Jason P. Gulbin,3 Allan G. Hahn,3 Alan H. Beggs,5
Simon Easteal,4 and Kathryn North1,2
1

Institute for Neuromuscular Research, Childrens Hospital at Westmead and 2Discipline of Paediatrics and Child Health, Faculty of Medicine,
University of Sydney, Sydney; 3Australian Institute of Sport and 4Human Genetics Group, John Curtin School of Medical Research, Australian
National University, Canberra; and 5Genetics Division, Childrens Hospital, Boston

There is increasing evidence for strong genetic influences on athletic performance and for an evolutionary tradeoff between performance traits for speed and endurance activities. We have recently demonstrated that the skeletalmuscle actin-binding protein a-actinin-3 is absent in 18% of healthy white individuals because of homozygosity
for a common stop-codon polymorphism in the ACTN3 gene, R577X. a-Actinin-3 is specifically expressed in fasttwitch myofibers responsible for generating force at high velocity. The absence of a disease phenotype secondary
to a-actinin-3 deficiency is likely due to compensation by the homologous protein, a-actinin-2. However, the high
degree of evolutionary conservation of ACTN3 suggests function(s) independent of ACTN2. Here, we demonstrate
highly significant associations between ACTN3 genotype and athletic performance. Both male and female elite
sprint athletes have significantly higher frequencies of the 577R allele than do controls. This suggests that the
presence of a-actinin-3 has a beneficial effect on the function of skeletal muscle in generating forceful contractions
at high velocity, and provides an evolutionary advantage because of increased sprint performance. There is also a
genotype effect in female sprint and endurance athletes, with higher than expected numbers of 577RX heterozygotes
among sprint athletes and lower than expected numbers among endurance athletes. The lack of a similar effect in
males suggests that the ACTN3 genotype affects athletic performance differently in males and females. The differential effects in sprint and endurance athletes suggests that the R577X polymorphism may have been maintained
in the human population by balancing natural selection.

The a-actinins are a family of actin-binding proteins

related to dystrophin. In humans, there are two genes
encoding skeletal-muscle a-actinins: ACTN2 (MIM
102573), which is expressed in all fibers, and ACTN3
(MIM 102574), which is restricted to fast (type 2) fibers.
The sarcomeric a-actinins are major components of the
Z line, where they crosslink actin thin filaments; they
likely perform a static function in maintaining the ordered
myofibrillar array, as well as a regulatory function in coordinating myofiber contraction (Blanchard et al. 1989;
Mills et al. 2001). We have recently demonstrated that
a-actinin-3 deficiency is common in the general population and is due to homozygosity for a premature stop
Received February 17, 2003; accepted for publication May 30,
2003; electronically published July 23, 2003.
Address for correspondence and reprints: Dr. Kathryn North, Institute for Neuromuscular Research, Childrens Hospital at Westmead,
Locked Bag 4001, Westmead, NSW 2145, Sydney, Australia. E-mail:
kathryn@chw.edu.au.
2003 by The American Society of Human Genetics. All rights reserved.
0002-9297/2003/7303-0015$15.00

codon in ACTN3 (R577X) (North et al. 1999). It is likely

that a-actinin-2 is able to compensate for the absence
of a-actinin-3 in type 2 fibers, although there is no upregulation of a-actinin-2 levels in response to a-actinin-3 deficiency (authors unpublished observations). However,
there is strong evidence to suggest that ACTN3 has been
maintained in the genome because of function(s) independent of ACTN2: ACTN3 sequence has remained
highly conserved, in evolutionary terms, since its divergence from ACTN2 1300 million years ago; a-actinin-2
and a-actinin-3 are differentially expressed, spatially and
temporally, during embryonic development; and ACTN2
expression does not completely overlap ACTN3 in mouse
postnatal skeletal muscle (Mills et al. 2001). In addition, the frequency of the a-actinin-3deficient genotype
(577XX) varies from 25% in Asian populations to !1%
in an African Bantu population; the frequency in Europeans is 18%. This raises the possibility that ACTN3
genotype confers differential fitness in humans, under certain environmental conditions. The force-generating ca627

628
pacity of type 2 muscle fibers at high velocity, the speed
and tempo of movements, and the capacity of the individual to adapt to exercise training are all strongly genetically influenced (Rankinen et al. 2002). Thus, we hypothesized that ACTN3 genotype may be one of the
factors that influence normal variation in muscle function.
Since any effect on muscle function will be most readily
observable at the extremes of human performance, we
collaborated with the Australian Institute of Sport to
study ACTN3 genotype frequencies in elite athletes.
We genotyped 436 unrelated white controls from
three different sources (150 blood donors, 71 healthy
children participating in an unrelated study, and 215
healthy adults participating in a talent-identification program with the Australian Institute of Sport), through use
of the genotyping methodology described by Mills et al.
(2001). Sex was known for 292 female controls and 134
male controls. We also genotyped 429 elite white athletes
from 14 different sports. Athletes were defined as elite
if they had represented Australia in their sport at the
international level; 50 of the athletes had competed in
Olympic Games. This study was approved by the institutional review boards of the Childrens Hospital at
Westmead, the University of Sydney, and the Australian
Institute of Sport.
Given the localization of a-actinin-3 in fast skeletalmuscle fibers, we hypothesized that deficiency of a-actinin-3 would reduce performance in sprint/power events
and would therefore be less frequent in elite sprint athletes. To test this hypothesis, we analyzed genotypes of
a subset of 107 elite athletes (72 male and 35 female)
in our cohort, classified a priori as specialist sprint/power
athletes by one of the authors (J.P.G.) at the Australian
Institute of Sport, blinded to genotyping results. This
group comprised 46 track athletes competing in events
of 800 m, 42 swimmers competing in events 200 m,
9 judo athletes, 7 short-distance track cyclists, and 3
speed skaters. For comparison, we analyzed a subset of
194 subjects (122 male and 72 female) classified independently as specialist endurance athletes, including 77
long-distance cyclists, 77 rowers, 18 swimmers competing over distances of 400 m, 15 track athletes competing in events of 5,000 m, and 7 cross-country skiers.
Thirty-two sprint athletes (25 male and 7 female) and
18 endurance athletes (12 male and 6 female) had competed at the Olympic level. Because of the stringency of
the classification criteria, 128 of our elite athletes could
not be unambiguously assigned into either the sprint/
power or endurance groups and were excluded from
subsequent analyses.
To test for homogeneity of ACTN3 allele and genotype frequencies between athlete and control groups, we
used the log-linear modeling approach described by Huttley and Wilson (2000), implemented in the statistical
programming language R (version 1.6.2), through use

Am. J. Hum. Genet. 73:627631, 2003

of the package hwde (contributed by J. Maindonald;

available from The R Project for Statistical Computing
Web site). x2 values were estimated using genotype numbers for comparisons between athletes and controls.
The genotypic profiles of the three control groups (150
blood donors, 71 healthy children, and 215 healthy
adults) did not differ significantly from one another
(x 2 p 0.19; P p .996) nor from a previously genotyped
group of 107 white Europeans (Mills et al. 2001), suggesting that the genotype frequencies in our control cohort are representative of the broader white population.
ACTN3 genotype frequencies did not vary significantly
between male and female control subjects, and, overall,
there was no significant deviation from Hardy-Weinberg
(H-W) equilibrium.
ACTN3 genotyping data from the control, sprint/
power, and endurance groups are summarized in table 1
and figure 1. There were no significant allele or genotype
frequency differences between the elite athlete group as a
whole and the controls. However, when the athletes were
divided into sprint/power and endurance groups and compared with controls, there was strong evidence of allele
2
p 23; P ! .001). There were
frequency variation (x[dfp5]
significant allele frequency differences between sprint ath2
letes and controls for both males (x[dfp1]
p 14.8; P !
2
.001) and females (x[dfp1] p 7.2; P ! .01). Sprint athletes
had a lower frequency of the XX (a-actinin-3 null) genotype (6% vs. 18%), and no female elite sprint athletes
or sprint Olympians were XX. The sprint athlete group
also had a higher frequency of the RR genotype (50%
vs. 30%) and a lower frequency of the heterozygous RX
genotype (45% vs. 52%), compared with controls. Elite
endurance athletes had a slightly higher frequency of the
XX genotype (24%) than did controls (18%). Importantly, allele frequencies in sprint and endurance athletes
Table 1
Number and Frequency (%) of ACTN3 Genotypes and Frequency
(%) of ACTN3 Alleles in Controls and Elite Sprint/Power and
Endurance Athletes

NO. (%)
GROUP (n)
Male:
Controls (134)
Sprint (72)
Endurance (122)
Female:
Controls (292)
Sprint (35)
Endurance (72)
Total:
Controls (436)
Sprint (107)
Endurance (194)

WITH

GENOTYPE

ALLELE
FREQUENCY
(%)

RR

RX

XX

40 (30)
38 (53)
34 (28)

73 (54)
28 (39)
63 (52)

21 (16)
6 (8)
25 (20)

57
72
54

43
28
46

88 (30)
15 (43)
26 (36)

147 (50)
20 (57)
25 (35)

57 (20)
0 (0)
21 (29)

55
71
53

45
29
47

130 (30)
53 (50)
60 (31)

226 (52)
48 (45)
88 (45)

80 (18)
6 (6)
46 (24)

56
72
54

44
28
46

Reports

629

Figure 1

ACTN3 genotype frequency in controls, elite sprint/power athletes, and endurance athletes. Compared with healthy white
controls, there is a marked reduction in the frequency of the ACTN3 577XX genotype (associated with a-actinin-3 deficiency) in elite white
sprint athletes; remarkably, none of the female sprint athletes or sprint athletes who had competed at the Olympic level (25 males and 7 females)
were a-actinin-3 deficient. Conversely, there is a trend toward an increase in the 577XX genotype in endurance athletes, although this association
reaches statistical significance only in females. Error bars indicate 95% CIs.

deviated in opposite directions and differed significantly

2
from each other in both males (x[dfp1]
p 13.3; P ! .001)
2
and females (x[dfp1] p 5.8; P ! .05). The differences between the two groups effectively canceled each other
out, explaining the lack of association when the entire
elite athlete cohort was compared with the control group.
Overall, there was also evidence of genotype variation
that is not explained by allele frequency differences
2
p 16.7; P ! .01). This suggested variation in H(x[dfp5]
W disequilibrium coefficients among groups, despite
there being no evidence for departure from H-W equilibrium overall. The effect was restricted to female sprint
2
2
(x[dfp1]
p 7.4; P ! .01) and endurance (x[dfp1]
p 6.0;
P ! .05) athletes, with more heterozygous female sprint
athletes than expected at H-W equilibrium (20 vs. 15)
and fewer than expected heterozygous female endurance
athletes (25 vs. 36). The allele-frequencyindependent
genotype differences between female sprint and endur2
ance athletes were highly significant (x[dfp1]
p 13.8;
P ! .001). No effect was seen in males, suggesting that
the effect of ACTN3 genotype on performance differs
between males and females.
Our findings suggest that the ACTN3 577R allele provides an advantage for power and sprint activities. No
female elite sprint athletes in our sample were a-actinin3 deficient (compared with 8% of males). In males, the
androgen hormone response to training is likely to make
a significant contribution to improvements in performance, so that the relative effect of a-actinin-3 on muscle
power may be reduced. Interestingly, all male Olympian
power athletes in our cohort had at least one copy of
the functional R allele of ACTN3 (associated with the

presence of a-actinin-3 in skeletal muscle), suggesting

that every variable counts at the highest levels of
sporting competition. Although at least 73 genetic loci
have been associated with fitness and performance phenotypes (Rankinen et al. 2002), ACTN3 is the first structural skeletal-muscle gene for which such an association
has been demonstrated.
The functional basis for this advantage is likely related
to the fact that a-actinin-3 is the predominant fast fiber
isoform in both mouse and human (Mills et al. 2001)
and may confer a greater capacity for the absorption or
transmission of force at the Z line during rapid contraction. Approximately 45% of the variation in fiber
type proportions is accounted for by genetic factors (Simoneau and Bouchard 1995). Sarcomeric a-actinins
bind to the gluconeogenic enzyme fructose-1,6-bisphosphatase (Gizak et al. 2003), to the glycogen phosphorylase amorphin (Chowrashi et al. 2002), and to the calsarcins (Frey et al. 2000; Frey and Olson 2002), which
interact with calcineurin, a signaling factor that plays a
role in the specification of muscle fiber type (Serrano et
al. 2001). Thus, a-actinin-3 may promote the formation
of fast-twitch fibers or alter glucose metabolism in response to training. In addition, a-actinin-3 may be evolutionarily optimized for the minimization of damage
caused by eccentric muscle contraction. The Z line in
fast, glycolytic fibers is the structure most vulnerable to
exercise-induced injury resulting in morphological damage and degradation of associated proteins, including
the a-actinins (Friden and Lieber 2001). We are currently
exploring the mechanism by which the presence or absence of a-actinin-3 alters muscle functionand, hence,

630
athletic performancethrough the generation and analysis of an Actn3 knockout mouse model.
From an evolutionary point of view, the challenge is to
explain the high frequency of the 577XX ACTN3 genotype, given the apparent power-performance advantage
of the 577RR genotype. One possibility is that the powerperformance effect of the 577RR genotype is only manifest in the extreme circumstances of athletic competition,
outside the range of normal human activity, and is consequently of minimal evolutionary significance. In this
model, the 577X allele could have been selectively neutral
during human evolution and become established in the
human population by random genetic drift. However, this
explanation is difficult to reconcile with the high level of
evolutionary conservation that we have previously demonstrated for ACTN3 (North et al. 1999; Mills et al.
2001).
It is also possible that the X allele is selectively neutral
but has reached its current frequency because of positive
selection on a beneficial polymorphism at a nearby locus
(i.e., genetic hitchhiking) (Kaplan et al. 1989). This
hypothetical variant would need to be in strong linkage
disequilibrium with 577X to result in the strong association observed in our study; however, it would be unlikely to reside within the ACTN3 gene itself, since the
577X polymorphism results in deficiency of the a-actinin-3 protein. The distance over which useful linkage disequilibrium extends varies considerably between
loci (Reich et al. 2001); however, a distance of 10 kb
has been proposed as a rough average value on the basis
of population modeling (Ardlie et al. 2002). The only
identified gene other than ACTN3 in the 20-kb region
centered on the R577X polymorphism is the CTSF gene
(MIM 603539), which encodes the papain-like cysteine
protease cathepsin F (Wang et al. 1998). The dbSNP
Home Page identifies nine polymorphic sites within the
CTSF gene, none of which alter the amino acid sequence
of the encoded protein. Furthermore, the only characterized function of cathepsin F is related to antigen processing in macrophages (Shi et al. 2000), making it an
unconvincing candidate for influencing athletic performance. Although we cannot completely rule out that variations in CTSF or in other, more distant genes have
influenced our results, it is more likely that the R577X
polymorphism is directly responsible for the observed
association with elite athletic performance.
The evolutionary model most consistent with our results is one in which 577XX genotype has been acted on
by positive natural selection. Our data demonstrate a
trend toward a higher frequency of the XX genotype in
endurance athletes, although the association reached statistical significance only in females; the frequency of the
577RX genotype was also lower in female elite endurance
athletes than in female controls (35% vs. 50%). However,
the effect may be stronger than is indicated by these re-

Am. J. Hum. Genet. 73:627631, 2003

sults, since a specific allelic association may be difficult

to detect in a heterogeneous cohort of mixed athletic disciplines. If the 577XX genotype enhances endurance performance as the 577R allele appears to enhance sprint
ability, then the 577R and 577X alleles may be maintained
in the population because they both confer selective advantages under different environmental conditions and
are thus kept at high population frequencies by balancing
selection. We are currently studying the frequency of alleles and the pattern of genetic polymorphisms flanking
the R577X locus in different human populations, to determine the origin of the X allele and to identify the form
and magnitude of any selective pressures that have acted
on the R577X locus.
Distinct beneficial effects on sprint and endurance athletic performance by different genotypes at a single locus
have also been observed in studies of the gene encoding
angiotensin-converting enzyme (ACE). The ACE gene
has two alleles, termed I and D; the I allele is associated with lower ACE activity in both serum and
tissue (Rieder et al. 1999). An increased frequency of
the ACE I allele has been observed in elite endurance
athletes (Gayagay et al. 1998; Montgomery et al. 1998;
Myerson et al. 1999; Nazarov et al. 2001). Conversely,
an increased frequency of the ACE D allele has been
associated with elite sprint performance (Myerson et al.
1999; Nazarov et al. 2001; Woods et al. 2001). The
absence of a correlation between the ACE I/D polymorphism and performance in other studies of elite athletes (Taylor et al. 1999; Rankinen et al. 2000) may be
explained by the failure of these studies to adequately
restrict their subjects to well-defined categories according to area of specialist performance, so that the allelic
association is canceled out (Nazarov et al. 2001).
It is likely that there is a trade-off between sprint
and endurance traits that imposes important constraints
on the evolution of physical performance in humans and
other vertebrates (Garland et al. 1990). This hypothesis
is supported by recent data from world-class decathletes,
which demonstrated that performance in the 100-m
sprint, shot put, long jump, and 110-m hurdles (which
rely on explosive power and fast fatigue-susceptible muscle fibers) is negatively correlated with performance in
the 1,500-m race (which requires endurance and fatigueresistant slow fiber activity) (Van Damme et al. 2002).
This suggests that an individual is inherently predisposed
toward specialist performance in one area (sprint/power
vs. endurance). In humans, this appears to have been
achieved, in part, through the maintenance of genetic
variation by balancing natural selection. The result is
that there are genetic differences among individuals, such
as we have demonstrated for the ACTN3 locus, that
may be useful predictors of athletic performance at the
elite level.

Reports

Acknowledgments
We are grateful to J. Maindonald, for assistance in implementing the hwde package in R, and to Professor Ron Trent
and Dr. Bing Yu, for provision of athlete DNA samples.

Electronic-Database Information
The URLs for data presented herein are as follows:
dbSNP Home Page, http://www.ncbi.nlm.nih.gov/SNP/index
.html
Online Mendelian Inheritance in Man (OMIM), http://www
.ncbi.nlm.nih.gov/Omim/ (for ACTN2, ACTN3, and CTSF)
R Project for Statistical Computing, The, http://www
.r-project.org/ (for the hwde package in R)

References
Ardlie KG, Kruglyak L, Seielstad M (2002) Patterns of linkage
disequilibrium in the human genome. Nat Rev Genet 3:
299309
Blanchard A, Ohanian V, Critchley D (1989) The structure
and function of a-actinin. J Muscle Res Cell Motil 10:
280289
Chowrashi P, Mittal B, Sanger JM, Sanger JW (2002) Amorphin is phosphorylase; phosphorylase is an alpha-actininbinding protein. Cell Motil Cytoskeleton 53:125135
Frey N, Olson EN (2002) Calsarcin-3, a novel skeletal musclespecific member of the calsarcin family, interacts with multiple Z-disc proteins. J Biol Chem 277:1399814004
Frey N, Richardson JA, Olson EN (2000) Calsarcins, a novel
family of sarcomeric calcineurin-binding proteins. Proc Natl
Acad Sci USA 97:1463214637
Friden J, Lieber RL (2001) Eccentric exercise-induced injuries
to contractile and cytoskeletal muscle fiber components.
Acta Physiol Scand 171:321326
Garland T Jr, Bennett AF, Daniels CB (1990) Heritability of
locomotor performance and its correlates in a natural population. Experientia 46:530533
Gayagay G, Yu B, Hambly B, Boston T, Hahn A, Celermajer
DS, Trent RJ (1998) Elite endurance athletes and the ACE
I allelethe role of genes in athletic performance. Hum Genet 103:4850
Gizak A, Rakus D, Dzugaj A (2003) Immunohistochemical
localization of human fructose-1,6-bisphosphatase in subcellular structures of myocytes. Histol Histopathol 18:135
142
Huttley GA, Wilson SR (2000) Testing for concordant equilibria
between population samples. Genetics 156:21272135
Kaplan NL, Hudson RR, Langley CH (1989) The hitchhiking
effect revisited. Genetics 123:887899
Mills M, Yang N, Weinberger R, vander Woude DL, Beggs
AH, Easteal S, North K (2001) Differential expression of
the actin-binding proteins, a-actinin-2 and -3, in different
species: implications for the evolution of functional redundancy. Hum Mol Genet 10:13351346
Montgomery H, Marshall R, Hemingway H, Myerson S, Clark-

631
son P, Dollery C, Hayward M, Holliman D, Jubb M, World
M, Thomas E, Brynes A, Saeed N, Barnard M, Bell J, Prasad
K, Rayson M, Talmud P, Humphries S (1998) Human gene
for physical performance. Nature 393:221222
Myerson S, Hemingway H, Budget R, Martin J, Humphries
S, Montgomery H (1999) Human angiotensin I-converting
enzyme gene and endurance performance. J Appl Physiol
87:13131316
Nazarov IB, Woods DR, Montgomery HE, Shneider OV, Kazakov VI, Tomilin NV, Rogozkin VA (2001) The angiotensin
converting enzyme I/D polymorphism in Russian athletes.
Eur J Hum Genet 9:797801
North KN, Yang N, Wattanasirichaigoon D, Mills M, Easteal
S, Beggs AH (1999) A common nonsense mutation results
in a-actinin-3 deficiency in the general population. Nat Genet 21:353354
Rankinen T, Perusse L, Raurama R, Rivera MA, Wolfarth B,
Bouchard C (2002) The human gene map for performance
and health-related fitness phenotypes: the 2001 update. Med
Sci Sports Exerc 34:12191233
Rankinen T, Wolfarth B, Simoneau JA, Maier-Lenz D, Rauramaa R, Rivera MA, Boulay MR, Chagnon YC, Perusse L,
Keul J, Bouchard C (2000) No association between the angiotensin-converting enzyme ID polymorphism and elite endurance athlete status. J Appl Physiol 88:15711575
Reich DE, Cargill M, Bolk S, Ireland J, Sabeti PC, Richter DJ,
Layery T, Kouyoumjian R, Farhadian SF, Ward R, Lander
ES (2001) Linkage disequilibrium in the human genome.
Nature 411:199204
Rieder MJ, Taylor SL, Clark AG, Nickerson DA (1999) Sequence variation in the human angiotensin converting enzyme. Nat Genet 22:5962
Serrano AL, Murgia M, Pallafacchina G, Calabria E, Coniglio
P, Lomo T, Schiaffino S (2001) Calcineurin controls nerve
activity-dependent specification of slow skeletal muscle fibers but not muscle growth. Proc Natl Acad Sci USA 98:
1310813113
Shi GP, Bryant RAR, Riese R, Verhelst S, Driessen C, Li Z,
Bromme D, Ploegh HL, Chapman HA (2000) Role for cathepsin F in invariant chain processing and major histocompatibility complex class II peptide loading by macrophages.
J Exp Med 191:11771186
Simoneau JA, Bouchard C (1995) Genetic determinism of fiber
type proportion in human skeletal muscle. FASEB J 9:
10911095
Taylor RR, Mamotte CDS, Fallon K, van Bockxmeer FM
(1999) Elite athletes and the gene for angiotensin-converting
enzyme. J Appl Physiol 87:10351037
Van Damme R, Wilson RS, Vanhooydonck B, Aerts P (2002)
Performance constraints in decathletes. Nature 415:755756
Wang B, Shi GP, Yao PM, Li Z, Chapman HA, Bromme D
(1998) Human cathepsin F: molecular cloning, functional
expression, tissue localization, and enzymatic characterization. J Biol Chem 273:3200032008
Woods D, Hickman M, Jamshidi Y, Brull D, Vassiliou V, Jones
A, Humphries S, Montgomery H (2001) Elite swimmers and
the D allele of the ACE I/D polymorphism. Hum Genet 108:
230232